Home Articles SAR Interferometry for DEM Generation

SAR Interferometry for DEM Generation

Manoj K Arora and Vinod Patel
Department of Civil Engg., IIT Roorkee, Roorkee

M L Sharma
Department of Earthquake Engg. IIT Roorkee, Roorkee
[email protected]

Remote sensing technology is gaining acceptance day by day for a variety of applications in various fields such as hydrology, geology, environment, transportation, ecology, earthquake engineering etc. The remote sensing systems may be categorised as active and passive systems. The active systems supply their own source of energy or illumination and use it to produce images of the earth surface, whereas the passive systems depend upon other source of energy (e.g., solar energy) for the purpose. There are a number of passive sensors onboard many remote sensing satellites that operate in the visible and infra red (optical) to microwave region of electromagnetic spectrum. Radar (Radio Detection and Ranging) is an active sensor operating in the microwave region. Among many radar-based systems, SAR, has commonly been used in remote sensing studies. The satellites such as ERS-1 and 2, RADARSAT and JERS carry SAR sensors.

The remote sensing data whether in optical or microwave region (active or passive) are used to prepare a variety of thematic maps at various scales with high degree of accuracy. Nevertheless, from engineering point of view, the third dimension (i.e., elevations or heights above mean sea level) of the features or objects on the earth’s surface, is of paramount importance. Various engineering activities such as environmental modelling, rainfall-runoff studies, landslide hazard zonation, seismic source modelling etc. require information about the third dimension in some form or the other.

Conventionally, height (altitude) has been determined through field surveys and to some extent from stereo-photogrammetry provided that the aerial photographs are available. The field surveys though provide elevation information to a high degree of accuracy, but are time consuming, laborious and costly, and provide information on point basis only. The point information on height may not be sufficient for conducting an engineering study on regional basis that requires spatial information. The spatial extent of height can be obtained from DEM. A DEM portrays the topographic information in the form of an array of numbers denoting location of features in terms of their x and y coordinates and the elevation. These data about the features are represented in raster form at a certain spatial resolution. Larger the resolution, cruder is the DEM representation and vice versa. Nevertheless, for plain areas where the elevations do not change much, the resolution may be kept large or coarse whereas in hilly areas, this is generally kept small or fine.

Traditionally, DEMs have been generated from contours in topographical maps produced from field surveys or aerial photogrammetry. The accuracy achieved from these DEMs may be limited to half the contour interval of the map. Thus, for example, a DEM produced from Survey of India (SOI) toposheet at 1:50,000 may be accurate up to a height accuracy of only 10m in plain areas. Due to the availability of stereo images from many remote sensing sensors operating in optical region (e.g., SPOT and IRS 1C/1D PAN), it is now possible to produce DEM at this level of accuracy and greater efficiency than before. Although, height accuracy in the range of 5 m to 10 m may be sufficient for many engineering projects, there are activities that require height information at centimeter to millimeter level accuracy. For example, monitoring of changes on the earth surface due to activities such as land subsidence, landslides, crustal deformations etc. Though, this monitoring can be achieved from Differential Global Positioning Systems (DGPS) but the information from these is again obtained on point basis only.

With the advent of InSAR, it may now be possible to obtain height information on spatial basis thereby producing DEM up to millimeter level accuracy. Due to this, the technology is gaining its momentum in many application areas such as lithospheric movements in geology, crustal deformation studies in seismology, global volcano monitoring, landslide monitoring, ice and glacial studies, etc. (Massonnet et al., 1996; Aldsorf and Smith, 1999; Kimura and Yamaguchi, 2000).

InSAR uses two SAR images acquired from satellites as mentioned above. Besides this, the data from a dedicated small test mission namely Shuttle Radar Topographic Mission for 15 days in February 2000 may also be used for InSAR. Moreover, with the expected launch of ENVISAT in near future, more SAR data shall be at the disposal of user community thereby strengthening the InSAR activities. Since, the SAR data are acquired in microwave region, it has an added advantage due to its capability to penetrate the atmosphere in virtually all weather conditions.

The aim of this paper is to provide an overview of aspects related to InSAR such as the concepts, the data acquisition and processing steps, and the availability of software to perform InSAR for DEM generation.

Sar Interferometry (InSAR)
Basically, InSAR can be categorised into three parts,

  • Single or simultaneous pass interferometry
    In this, two images are simultaneously acquired from two antennas separated by a distance known as baseline. The SRTM is based on this principle where receiving antennas are located at two ends of a boom of 60 meters as baseline length.
  • Repeat or dual pass interferometry
    It is also known as repeat track interferometry. Here, two images of the same area are taken in different passes of satellite. The SAR data acquired from satellites namely ERS-1 and ERS-2, JERS-1 and RADARSAT may be used in this fashion to produce SAR interferograms (Rao and Rao, 1999).
  • Three pass interferometry or differential InSAR

Here, three SAR images of the same area are acquired in three different passes to generate two interferograms. The arithmetic difference of the two interferograms is used to produce a differential interferogram. From this differential interferogram, a millimeter level accuracy in elevation measurement can be achieved.

Fig. 2: A depiction of fringes
(Lines of constant colour
correspond to constant phase
difference and therefore
constant terrain altitude; i.e.
height contours)
(Source:

Sar Interferometric Processing

Geometry of InSAR
The geometry of InSAR may be understood with the help of Fig. 1. In this figure, A1 and A2 are the two radar antennas that simultaneously view the same surface and are separated by a baseline vector B with length B and angle á with respect to horizontal. A1 is located at height h above some datum. The distance between A1 and the point on the ground being imaged is the range , while is the distance between A2 and the same point. The aim is to determine the elevation h of each point in the image. The topography z(y) can be inferred from the phase measurement to a precision of several meters, assuming that the 2 ambiguity inherent in any phase measurement can be solved (Equation 1).

Z (y) = h – rcosq    (1)

where q is the look angle of the radar.

A SAR interferogram, viewed as a fringe pattern, shows the relative difference between phases of two images. The phase difference depends on the geometry of the two antenna tracks and the image point and thus is proportional to the difference in path delays from two antennas and is given by,

j = 4p (r1 – (r + d r) / l    (2)

where l is the wavelength.

To determine h, the interferometric processing steps that are generally followed can now be enumerated as:

  • Selection of suitable pair of SAR images
  • Geometric registration
  • Interferogram generation
  • Phase unwrapping
  • Extraction of elevations from phases

Selection of SAR images
The first requirement is the availability of two SAR images in complex form. Complex SAR data refer to a set of data that has a real (cosine) and an imaginary (sine) component. The two values combine as vectors to provide the overall phase and intensity of a wave. Both these components of backscattered signals are measured by the SAR sensor onboard the satellite. This provides two resulting data streams, namely ‘I’ (representing In-phase / intensity / cosine component) and ‘Q’ (representing Quadrature / phase / sine component) data streams (https://www/asf/alaska.edu/).

The selection of the images is made on the basis of baseline length and the time period between two image acquisitions. Depending upon the application and the spatial resolution of the data, the baseline length can be chosen. For example, in the case of ERS–1 and 2, the baseline may be taken as 150 to 300m for topographic applications, 30 to 50m for surface change detection and up to 5m for surface feature movement studies such as crustal deformations, lithospheric movements, movement of glaciers etc. Also, the time gap between two passes of satellite may not be kept large as there may be some changes in the scene that may lead to temporal de-correlation. However, the temporal de-correlation, in the case of ERS-1 and 2 may be taken care of by tandem operation of two satellites at a small temporal resolution of as low as one day (Rao and Rao, 1999).

The selected raw data are then processed to convert SAR signals to image products like, Single Look Complex and GTC Geocoded Terrain Corrected with the help of DEM. This processing requires knowledge about the precise orbit and calibration parameters such as time reference and intervals of each image, and the chosen spatial and temporal resolutions of the images.
Geometric registration
Registration of SAR images (both intensity and phase) is generally carried out at sub-pixel level for accurate results. At the first instance, the registration at pixel level is accomplished by a simple two-dimensional transformation using a limited number of Ground Control Points identified in the two images. This is followed with a sub-pixel registration using a large number of GCPs (of the order of 40 to 50). In the first attempt, two images are divided into sub-areas. Then, the coordinates of each sub-area are determined according to one of the following correlation method (Rao and Rao, 1999):

  • Cross correlation of pixel intensities,
  • Maximum value of the coherence coefficient,
  • Evaluation of the maximum intensity, and
  • Minimum of the average fluctuation of the phase difference image.

Interferogram generation
Two SAR images are combined to produce a SAR Interferogram to reveal information about the third dimension (elevation) of the object and to measure small displacements of objects between the two image acquisitions. An interferogram is an image acquired by making the phases of two SAR images of the same terrain to interfere.

Thus, after registration, the complex interferograms are formed by multiplying each complex pixel of the first image by the complex conjugate of the same pixel in the second image. The interferogram thus generated is a complex image itself. The intensity of the interferogram is a measure of cross correlation of the images. A careful observation of the fringes in Fig. 2 reveals that closer are the fringes, more are the topographical changes or height variations.

Phase unwrapping
As the height of the terrain increases, the phase also increases steadily. Since phase values are periodic functions of 2p, they automatically get wrapped after reaching 2p, which is not a desirable situation. Phase unwrapping is a technique that permits retrieving the unwrapped phase from the wrapped phase, which for the InSAR, is a necessary step for the generation of DEM (Fornaro et al., 1996). There are many methods of phase unwrapping.

Phase to height conversion
As a final step, the terrain height may be determined using several methods which convert phases into terrain heights (Bilrgmann et al., 1999). Some of them are:

  • Normal baseline method
  • Integrated incidence angle method
  • Baseline rotation method.

Normal baseline method operates on unwrapped phase where change in the height of the terrain is related to change in the phase using the following equation (Rao and Rao, 1999):

Dh = (lr1 sinq / 4pB’) Dj    (3)

where r1 is the range in the first image,
B’ is the normal baseline,
l is the wavelength,
Dj is change in phase, and
q is the incident angle.

Thus, utilising the above-mentioned steps, the relationship between the heights and the phase differences can be determined. These heights determined for each pixel of the SAR image form the necessary DEM in raster form. The steps, though they look, simple require complex mathematical algorithms at each step and therefore, it is desirable that the processing be done using some software. In the next section, some of the softwares for InSAR have been listed.

Software for InSAR
A number of software packages have been developed to process and analyse SAR interferometric data. The early days of InSAR saw the development of packages by many research institutes. Later, a few commercial packages have also become available. The salient features of some of the commercial and non-commercial packages have been given in Table 1. It provides information about the supported sensors and formats as well as the other generated products as mentioned before. It can be seen that most of the supported sensors are from space borne systems. Among them ERS-1 and 2 imageries are supported by all the packages. It may also be noticed that most of the packages can be used for the generation of the DEM. To represent the DEM as a final interferogram product, additional information about map projection, grid size, etc. may also be required. A log file can also be generated from most of these software packages that contains a large amount of information about the parameters used for processing as well as its performance.

Conclusions
An overview of various aspects related to InSAR such as the concepts, the data acquisition and processing steps, and the availability of software to perform InSAR are described for DEM generation.

Following general conclusions can be drawn:

  • In less than ten years, InSAR has demonstrated its capabilities for quantitative measurements of surface topography.
  • InSAR has the potential of providing DEM at an accuracy of 1-10 cm, which can further be improved to millimeter level from Differential InSAR.
  • SRTM and ENVISAT are the dedicated missions for providing necessary data for InSAR. However, the ERS-1, ERS-2, JERS-1 and RADARSAT can also be used for interferometric purposes.
  • Though many softwares have been developed, some aspects regarding data format and data quality need to be standardised in them to get uniform end products.
  • By and large, InSAR can effectively be applied in many areas that are concerned with mapping and monitoring of earth’s surface at higher accuracy.

References
Asdorf, D. and Smith L. C., 1999, Interformetric SAR observations of ice topography and velocity changes related to the 1996 Gjalp sub-glacial eruption Iceland, International Journal of Remote Sensing, 20, 3031-3050.

  • Bilrgmann R., Rosen P. A., and Fielding E. J., 1999, SAR Interferometry to measure earth’s surface topography and its deformation, 1-57.
  • Fornaro G., Franceschetti G., and Lanari R., 1996, Interferometric SAR phase unwrapping using Green’s formulation, IEEE Transactions on Geoscience and Remote Sensing, 34, 720-727.
  • Gens R., 1999, SAR Interferometry: software, data format and data quality, Photogrammetric Engineering and Remote Sensing, 65, 1375-1378. https://www.asf.alaska.edu/, 2000, SAR frequently asked questions, https://earth.esa.int/, 1995, SAR Interferometry with ERS
  • Kimura H., and Yamaguchi, Y., 2000, Detection of landslide areas using satellite radar interferometry, Photogrammetric Engineering and Remote Sensing, 66, 337-344.
  • Massonnet D., Feigl K. L., Rossi M., and Vadon H., 1996, Co seismic deformation field of the M 6.7 Northridge, California Earthquake of January 17, (1994), Recorded by Two Radar Satellites using Interferometry”, Geophysical Reaserch Letters, Vol. 23, No. 9, 969-972.
  • Rao K. S. and Rao Y. S., (1999), “Seminar on Recent Developments in Differential SAR INTERFEROMETRY and its Applications”, Lecture Notes, IIT Bombay, 1-11.